Hi everybody: I *think* I want to whitelist an address, but stop SA from autolearning the whitelisted mail as ham.
Before I get started explaing why, i realize this topic has been discussed here. This thread is enlightening: http://article.gmane.org/gmane.mail.spam.spamassassin.general/29088 Here's a short take of the thread: Simon argues that this is probably not a feature we need ("a solution to a problem that doesn't exist," i think Simon says). Carlo counters that mail from the sa-talk list is a great example of mail that requires just such a feature. For example, Carl argues, much of the mail on sa-talk contains snippets of spam email, therefore making it an excellent candidate for a whitelist (because it may be classified as spam by the filter); while at the same time, one would not want the "spammy" tokens from sa-talk emails added to one's ham database, thereby poisoning the ham. Simon argues back that one doesn't need to whitelist mail from the sa-talk list anyway, because spamassassin won't classify sa-talk mail as spam anyway. He asks (i paraphrase), "is anybody having a problem with this?" ..Well, I wasn't, until today. Three messages from sa-talk were classified as spam by my SpamAssassin 2.60 installation today, specifically messages from the "Mailing lists and compliance verbage" thread. (scoring from SA attached below). So, when I saw this, I said, "well, I should whitelist-to the sa-talk list address," which I did, but then I quickly saw all subsequent mail was getting marked as autolearn-ham, which I quickly realized would be a bad thing. So I guess, confidential to Simon: yeah, I appear to be having a problem with this (admittedly minor--I've been on the list for a while and this is the first time this has happened to my knowledge). It appears the feature Carlo talks about here makes sense. In fact, I'd argue one very rarely would want whitelist mail marked as autolearn-ham. I mean, since the thing is whitelisted anyway, we don't really need to tell bayes about it, right? Or maybe so...maybe someone can enlighten me here. Either way, I agree with Carlo on this one: it'd be nice to be able to both whitelist an address but to stop the filter from autolearning from that mail. Regards; DaC (Attached SA scoring--this is for the message from Fred at email "tech2 at i-is dot com", subject "Mailing lists and compliance verbage", sent "Fri, 19 Dec 2003 14:29:39 -0500") Content analysis details: (6.7 points, 6.0 required) pts rule name description ---- ---------------------- -------------------------------------------------- 2.7 SENT_IN_COMPLIANCE BODY: Claims compliance with spam regulations 1.7 NO_COST BODY: No such thing as a free lunch (3) 2.6 OPT_IN_CAPS BODY: Talks about opting in (capitalized version) 2.7 BILL_1618 BODY: Claims compliance with Senate Bill 1618 1.8 CANNOT_BE_SPAM BODY: Claims "cannot be considered spam" -4.9 BAYES_00 BODY: Bayesian spam probability is 0 to 1% [score: 0.0000] 0.1 CLICK_BELOW Asks you to click below ------------------------------------------------------- This SF.net email is sponsored by: IBM Linux Tutorials. Become an expert in LINUX or just sharpen your skills. Sign up for IBM's Free Linux Tutorials. Learn everything from the bash shell to sys admin. Click now! http://ads.osdn.com/?ad_id=1278&alloc_id=3371&op=click _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
